新疆地区地震发生的统计分析
|
摘要
根据新疆地区本世纪的地震记载，分别取５．５、６．０、６．５、６、７．０、７．２为震级下限，视各组资料为更新过程的样本、分析地震发生的统计特征．经统计检验，不同震级下限的地震发生皆可认为是泊松过程；本世纪初中强地震在记载中可能有遗漏，为避免由此引出的干扰，使用Ｍｓ≥６．５的资料不应早于１９１４年，使用Ｍｓ≥５．５与Ｍｓ≥６．０的资料均应从１９３１年开始；在地震发生大体上为油松过程的情况下，地震间隔时间并不匀称地切合指数分布。在某些区间上表现出显著的丛聚性，其原因还应探索，它对地震预测有一定的参考作用；用累积概率表示发震的可能程度可作为马氏链预测发震时间的补充；Ｍｓ≥６．０、，Ｍｓ≥６．５的间隔时间同时可认为服从对数正态分布；新疆地区中强以上地震中≥Ｍ级的地震平均复现期 与Ｍ有极显著的线性相关性，以此关系可算出该地区近百年来平均的ｂ值为０．７４１；新疆地区７级以上地震的近３０年的呼应与在太阳黑子活动谷年附近的多发以及７．５级以上大震的活动等都形成小概率事件，应探索缘由，给予适当解决；对小样本用６３．２％分位点值为λ－１的估计值和通过λ（Ｍ）计算长期平均ｂ的方法有较好地统计效应。
According to the historical recordings of earthquakes in Xinjiang since this century,the author analyzed the earthquake occurrence patterns,making 5. 5, 6. 0, 6. 5, 6 , 7. 0, 7.2 the lower bounds of magnitude and each group of data the update processing samples. Statistical examination showed that earthquakes with different magnitude limitation can be regarded as Poisson process. Because of the possibility of strong earthquakes'recording lackage at the beginning of the century,timp period for shock events is suggested in the paper in order to avoid unnecessary interference,that is,seismic data of Ms≥6.5 are to be taken after 1914, while those of Ms≥5. 5 and 6. 0 from 1931 will be available. Acturally,earthquake intervals do not coincide with exponential distributions in spite of their Poisson process belongings.Sometimes earthquakes occurred in cluster,which may be of some use in earthquake prediction. Using cumulative probability to estimate the possible stage of earthquake occurrence can fulfill the origin time of Markov chain. Time intervals of earthquakes with Ms≥6. 0 and Ms≥ 6. 5 are supposed to obey lognormal distribution. Since there is a close linear correlation be 0tween the mean recurrence of earthquake ho of magnitude ≥M and those of magnitude= M, we work out that their mean values b are 0. 741 in the region for nearly one hundred years. Earthquakes with M≥7. 0 in Xinjiang in 30 years that corresponded with the multiple events of sunspot activity in valley year as well as the large earthquakes (Ms≥7. 5) activity are to be considered minor probability events,which we think need further study and suitable explanation. In the paper,better statistical effects on small samples obtained by 63. 2 % fractile value λ-1estimation and long period mean value b calculated through λ(M) computation are presented.
According to the historical recordings of earthquakes in Xinjiang since this century,the author analyzed the earthquake occurrence patterns,making 5. 5, 6. 0, 6. 5, 6 , 7. 0, 7.2 the lower bounds of magnitude and each group of data the update processing samples. Statistical examination showed that earthquakes with different magnitude limitation can be regarded as Poisson process. Because of the possibility of strong earthquakes'recording lackage at the beginning of the century,timp period for shock events is suggested in the paper in order to avoid unnecessary interference,that is,seismic data of Ms≥6.5 are to be taken after 1914, while those of Ms≥5. 5 and 6. 0 from 1931 will be available. Acturally,earthquake intervals do not coincide with exponential distributions in spite of their Poisson process belongings.Sometimes earthquakes occurred in cluster,which may be of some use in earthquake prediction. Using cumulative probability to estimate the possible stage of earthquake occurrence can fulfill the origin time of Markov chain. Time intervals of earthquakes with Ms≥6. 0 and Ms≥ 6. 5 are supposed to obey lognormal distribution. Since there is a close linear correlation be 0tween the mean recurrence of earthquake ho of magnitude ≥M and those of magnitude= M, we work out that their mean values b are 0. 741 in the region for nearly one hundred years. Earthquakes with M≥7. 0 in Xinjiang in 30 years that corresponded with the multiple events of sunspot activity in valley year as well as the large earthquakes (Ms≥7. 5) activity are to be considered minor probability events,which we think need further study and suitable explanation. In the paper,better statistical effects on small samples obtained by 63. 2 % fractile value λ-1estimation and long period mean value b calculated through λ(M) computation are presented.
引文
1陈希孺，等．非参数统计．科学出版社，1989．2M·J·纳特雷拉．实验统计学（毛镇道等主译），1990．3S·希格尔．非参数统计（北星译）．科学出版社，1986．4陈家合．应用概率论．科学出版社，1992．5U·S·NationalBureauofStandards．TableoftheBinomialProbabilityDistribution．1950．6常用数理统计表．科学出版社，1974．7陆运忠，等．地震预报的地震学方法．地震出版社，1985．8戴树森，等．可靠性试验及其统计分析（上）．国防出版社．1983．
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心